Skip to content

Non-destructive Assessment of Tungsten Carbide-Backed PCD-Steel Brazed Joints Using Active Infrared Thermography

FieldValue
JournalJournal of Materials Engineering and Performance
Published2026-02-26
AuthorsMartin Kahlmeyer, Ammar Ahsan, Christian Wolf, Igor Kryukov, Eugen Prints
DOI10.1007/s11665-026-13519-5

Induction-excited infrared thermography separated defect-free and artificially defective PCD-to-steel brazed joints through their heating responses. Defect-free joints averaged 3.5 kN in push-off shear testing, while every defect group lost more than half that load.

The tested cutting element was a mixed-grain PCD saw tooth with a WC-8%Co backing. Its PCD layer combined 30 μm and 2 μm diamond grains. The teeth were cut from a disc by wire EDM, sandblasted, and ultrasonically cleaned in ethanol for 5 minutes. The parts remained in ethanol until flux application.

Steel seats were milled with a carbide tool. Mean carbide roughness was Ra 2.2 μm, with 0.15 μm standard deviation. Steel roughness was Ra 1.5 μm, with 0.1 μm standard deviation.

The filler was Ag449 under ISO 17672:2024, containing 49% Ag, 16% Cu, 23% Zn, 7.5% Mn, and 4.5% Ni. Preplaced filler foil was 0.1 mm thick and had a melting range of 680 to 705 °C.

Brazing occurred in ambient atmosphere with a 20 kHz, 10 kW generator and a one-sided coaxial coil. The setpoint was 720 °C, measured by pyrometry at the carbide side of the interface. The coil stood 1 mm from the saw band. The SCARA placement system reached 0.02 mm positional accuracy and 1° angular accuracy.

The study compared one reference condition with three artificial defect conditions. Each induction cycle had a heating phase and a holding phase. Phase entries list time, PWM, and frequency. PWM is expressed in per mille, as specified by the generator manufacturer.

ConditionArtificial defectPhase 1: time, PWM, frequencyPhase 2: time, PWM, frequency
C1None, reference12 s, 600‰, 20 kHz3 s, 500‰, 20 kHz
C2Insufficient temperature9 s, 600‰, 20 kHz1.5 s, 500‰, 20 kHz
C3SAE 304 foil between tip and filler, 0.05 mm thick and 1 mm wide12 s, 600‰, 20 kHz3 s, 500‰, 20 kHz
C40.5 mm air gap between tooth and seat12 s, 600‰, 20 kHz3 s, 500‰, 20 kHz

C2 was classified as insufficient temperature and used shorter heating and holding phases. All PCD tips had a thickness of 2 mm.

The FLIR SC5600-M used a Stirling-cooled InSb detector with 640 × 512 pixels. The spectral range was 2.5 to 5.1 μm, and the noise-equivalent temperature difference was below 25 mK. Imaging used a 159 × 127 pixel window at 200 Hz. After camera software scaling, analyzed frames typically measured about 200 × 350 pixels.

An induction coil reheated the tips at 20 kHz and 250‰ PWM for 2 seconds. Heating stopped after 2 seconds, and the camera recorded the first 5 seconds of cooling. The total recording lasted 7 seconds. The camera stood 410 mm from the near end of the saw tips. The coil and camera used a transmission arrangement, with each on an opposite side.

Steel heated faster than carbide because steel is ferromagnetic. Heat then passed from steel through filler and carbide into PCD. PCD had about seven times the thermal conductivity of the carbide layer. Poor wetting, gaps, and inserted foil added thermal resistance. These defects delayed the PCD response during heating.

Visual inspection showed reduced filler spreading in C2 and the continuous C4 gap. The thin C3 foil was not obvious because the joint looked similar to C1.

Micro-computed tomography used 140 kV, 10 W, and an approximately 13 μm voxel size. It collected 1601 projections during a 360° rotation. C1 had a continuous filler layer with small isolated pores. C2 showed incomplete filling. Micro-CT resolved C3 as separation at the foil position and C4 as large continuous voids.

SEM at 30 kV showed local unbonded regions in C2 that appeared as cracks. SEM also confirmed the inserted foil and large unfilled regions. The authors associated the C2 unbonding with likely insufficient filler melting.

Visual inspection and SEM alone did not expose every defect. C2 bonding was difficult to identify in SEM, while C3 was difficult to identify visually. The mechanical test exposed all three defect types.

The PCD’s homogeneous dark surface was well suited to thermographic measurement. A line object along the PCD layer averaged IR intensity at each time step. Each curve used the mean of five measurements. The curves were plotted in double-logarithmic form to expose deviations during heating.

Change in IR intensity, DI, represented the temporal signal change at each pixel or region. The reference C1 heated the PCD fastest. C2 produced a lower PCD heating rate because bonding was poor. C3 also reduced the heating rate relative to the silver-based filler. The authors attributed this reduction to the stainless steel foil’s lower conductivity. C4 caused the largest delay because the air gap formed a macroscopic thermal barrier.

Single frames did not show clear defect contrast. The defects were small relative to the heated area, and induction created a dominant global thermal field. Temporal curves therefore provided greater sensitivity than individual thermograms.

The authors suggest that local intensity profiles at 1 second could support thresholding and defect typing. The paper did not define production thresholds or classification rules.

Push-off testing pushed each tooth from the carbide side while a clamp held the steel carrier. C2 averaged 1.6 kN, C3 averaged 1.8 kN, and C4 averaged 1.0 kN. The separation of all four thermal responses agreed with the destructive results. This supports thermography as a joint-quality indicator, although the study reports no correlation coefficient or statistical classification performance.

Unlike push-off testing, thermography left each joint available for later mechanical validation. Earlier studies used acoustic emission, electrical resistance, ultrasonics, eddy currents, X-ray micro-CT, thermal probes, modal analysis, and pulsed eddy current. The authors describe their test as the first thermography application to induction-brazed cutting elements during brazing.

The automated acquisition and analysis support the authors’ proposal for 100% inspection. The result was a feasibility demonstration, not a validated production acceptance test. The study used one bandsaw geometry and one mixed-grain PCD architecture.

No false-positive rate, false-negative rate, repeatability limit, or cycle-time target was reported. The authors state that industrial evaluation methods still need development. They also propose moving thermography into the brazing process because both operations use induction. That in-process method was proposed rather than demonstrated on a production station.

The paper did not report long-term effects of repeated reheating on the PCD or filler. Surface finish, PCD grade, and tooth geometry were not varied. These gaps limit direct transfer of the reported responses to other tool geometries or diamond architectures.

Brazing quality strongly affected insert retention in the tested joints. An interface defect changed the PCD heating response and reduced the push-off load by more than half. A thin foreign-metal layer was also difficult to identify by visual inspection.

The authors propose a qualified reference joint as the baseline for each time-response comparison. Evaluation should use a defined PCD region because single images did not separate the defects. Both joining and inspection use induction heating, which supports integration during the joining process.

The study does not rank PCD grades because it used only one mixed-grain architecture. For this joint architecture, the results support qualifying the PCD insert, WC-Co backing, filler, and brazing parameters together.

View Original Abstract

Abstract Brazing is the most commonly used method for joining carbide and carbide-coated polycrystalline diamond inserts to steel bodies of metal and wood saw blades. However, there is currently no non-destructive inline method for evaluating joint quality. This may lead to insufficient tool performance and high costs, particularly for large saw blades with many teeth. To address this challenge, we present a novel in-line testing method designed to reliably and efficiently assess brazing quality. Polycrystalline diamond inserts with carbide backing were inductively brazed to steel surfaces using a silver-based brazing alloy. Artificial defects, including incorrect positioning, inclusions, or deliberately chosen incorrect brazing parameters, were introduced into some of the brazed joints. The heat transfer rate during brazing was monitored through thermography. The proposed method reliably identified artificially introduced defects in brazed joints and showed good correlation with destructive push-off shear testing. These findings demonstrate the potential of induction-excited thermography for inline quality assurance of brazed joints in industrial saw blade production.

  1. 2013 - Advances in Brazing: Science Technology and Applications
  2. 2017 - Die Elektrische Widerstandsmessung Zur Zerstörungsfreien Prüfung Von Lötnähten Am Beispiel Von Hartmetall-Stahl-Verbunden: (English: The Electrical Resistance Measurement for Non-Destructive Testing of Brazed Carbide—Steel Joints.) [Crossref]